FIELD AND FORAGE CROPS Farming Practices Influence Wild Populations on Squash and Pumpkin

1 2 3 RACHEL E. SHULER, T’AI H. ROULSTON, AND GRACE E. FARRIS

Department of Environmental Sciences, University of Virginia, Charlottesville, VA 22904-4123

J. Econ. Entomol. 98(3): 790Ð795 (2005) ABSTRACT Recent declines in managed honey , Apis mellifera L., colonies have increased interest in the current and potential contribution of wild bee populations to the pollination of agricultural crops. Because wild often live in agricultural Þelds, their population density and contribution to crop pollination may be inßuenced by farming practices, especially those used to reduce the populations of other . We took a census of of squash and pumpkin at 25 farms in Virginia, West Virginia, and Maryland to see whether pollinator abundance was related to farming practices. The main pollinators were pruinosa Say; honey bees, and bumble bees (Bombus spp.). The was the most abundant pollinator on squash and pumpkin, occurring at 23 of 25 farms in population densities that were commonly several times higher than that of other pollinators. Squash bee density was related to tillage practices: no-tillage farms hosted three times as great a density of squash bees as tilled farms. Pollinator density was not related to use. Honey bee density on squash and pumpkin was not related to the presence of managed honey bee colonies on farms. Farms with colonies did not have more honey bees per ßower than farms that did not keep honey bees, probably reßecting the lack of afÞnity of honey bees for these crops. Future research should examine the economic impacts of managing farms in ways that promote pollinators, particularly pollinators of crops that are not well served by managed honey bee colonies.

KEY WORDS pollination, squash, pumpkin, sustainable agriculture, tillage

INSECTS ARE OFTEN VIEWED as the scourge of agriculture, 1998). The prospect of future honey bee shortages has yet many food crops require pollination to set led to a recent interest in the role of wild pollinators fruit. Thus, farm management practices must attempt in agricultural systems (Allen-Wardell et al. 1998, Kre- to reduce the negative effects of herbivorous or dis- men and Ricketts 2000, Westerkamp and Gottsberger ease-transmitting insects while maintaining an envi- 2000, Kremen et al. 2002). Because wild pollinators ronment conducive to pollinator activity. The honey generally cannot be introduced suddenly to agricul- bee, Apis mellifera L., is the predominant managed tural systems in adequate numbers to ensure pollina- pollinator in much of the world (Robinson et al. 1989). tion, successful management approaches are likely to Because it occurs in very large colonies, visits many focus on managing farm conditions rather than the different crops, and can be transported into and out of pollinators themselves. agricultural Þelds, it has provided agriculture with the Pioneering work by Kremen et al. (2002) has shown ability to take aggressive insect control measures that wild bee populations vary with farming practices through much of the growing season without suffering and the distance from farms to natural habitats. Work- substantial losses of insect pollination. ing in a major agricultural area of California, they The number of managed honey bee colonies in the showed that organic farms near natural habitats hosted United States has recently declined due to difÞculties sufÞcient wild bees to provide full pollination services in managing them. These difÞculties include the re- for watermelon (Citrullus spp. Shrad.), a lucrative cent establishment of parasitic mites and hybridiza- crop with large pollination requirements. Wild bee tion with the Africanized honey bee, Apis mellifera populations were diminished at all other farms, and scutellata (Ruttner), in some regions (Peng and Nasr full pollination required the addition of honey bees. 1985, Weinberg and Madel 1985, Allen-Wardell et al. This work points to the possibility that farm manage- ment practices that encourage wild pollinator popu- 1 Department of Biology, Oberlin College, Oberlin, OH 44074. lations may provide ensurance against pollination 2 Corresponding author: University of Virginia, Blandy Experimen- losses incurred by further honey bee declines and tal Farm, 400 Blandy Farm Lane, Boyce, VI 22620 (e-mail: reduce costs associated with renting or maintaining [email protected]). 3 Department of Hispanic Studies, Brown University, Providence, honey bee colonies when they are unnecessary. Some RI 02903. historical evidence shows that honey bees became

0022-0493/05/0790Ð0795$04.00/0 ᭧ 2005 Entomological Society of America June 2005 SHULER ET AL.: SQUASH AND PUMPKIN POLLINATION 791

Fig. 1. Distribution of 25 participating farms that grew squash, pumpkin, or both. increasingly necessary in North American agriculture pumpkin is the bee Peponapis pruinosa Say, a special- when intensive farming practices reduced the popu- ized, widespread pollinator that collects only lations of wild local bee populations (Batra 1995). from the genus (Hurd et al. 1974). Because To sustain populations, bee species require food there are no wild Cucurbita in this region, the P. resources throughout their active period and undis- pruinosa population is entirely dependent on culti- turbed nesting substrate during their developmental vated Cucurbita and cannot maintain refuge popula- period. Food resources comprise pollen and nectar, tions far from agricultural areas. Our work provides an which together provide the protein, carbohydrates, indication of which bees are primarily responsible for and micronutrients required for larval development pollination of squash and pumpkin in this region and and adult maintenance (Michener 2000). Nesting sub- which farming practices may have the greatest inßu- strates vary, but most bee species are either cavity ence on their pollinator populations. We focus on nesters that occupy existing structures such as hollow farming practices that seem most directly related to plant stems (Frankie et al. 1998) or ground nesters that the life cycle of wild bees: tillage (survival of immature excavate tunnel systems in earthen banks or bare bees), crop diversity (continual food supply), and patches of soil (Chapman et al. 1990). Natural cavities pesticide use (direct impact on adults). This study has are most likely to occur outside the planting area, but implications both for the economics of agriculture and bare earth occurs commonly within Þelds, and many the conservation of biodiversity in agroecosystems. bee species nest alongside crops (Mathewson 1968). The survival of offspring within planting areas de- pends on nests not being disturbed during develop- Materials and Methods ment, which takes only a few weeks during the sum- mer in species that have multiple generations Participating Farms. We compiled a study group of (multivoltine) but takes most of the year for univol- 25 farms within an Ϸ100 by 130-km area of Virginia, tine species and for the overwintering generation of West Virginia, and Maryland (Fig. 1). Participating multivoltine species. farmers were initially contacted at regional farmersÕ In the current study, we examine the effect of farm- markets or at their own farms by driving through the ing practices on pollinator populations of cultivated countryside looking for large plantings of squash and squash and pumpkin in the tristate border area of pumpkin. Farmers were interviewed concerning their Virginia, West Virginia, and Maryland. Squash and management practices, including pesticide use, tillage, pumpkin (both in genus Cucurbita) are valuable, com- diversity of crops grown, use of managed honey bee monly grown crops that require insects for pollination. colonies, and the number of consecutive years that Although honey bee colonies are often placed in squash or pumpkin had been grown on the site. squash and pumpkin Þelds for pollination, honey bees Twelve of the participating farms did not use pesti- prefer other crops, weeds, and wild plant species and cides, whereas 13 applied one or more types of pes- often fail to visit the target plants if other options are ticides. Planting area of the target crops ranged from available (Delaplane and Mayer 2000). One of the Ͻ0.5 to 40 ha (median 0.8). Total farm area ranged most effective and persistent pollinators of squash and from Ͻ0.5 to 400 ha (median 80.9). 792 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 98, no. 3

Insect Surveys. Each farm was surveyed for insect 0.5 visitation during 1 d between 7 July and 5 August 2003. p = 0.006 Surveys were limited to sunny-to-moderately cloudy 0.4 days in the morning. Squash and pumpkin ßowers are open from predawn until Ϸ1000 hours in this part of the United States, and individual ßowers last a single 0.3 day only. Although squash bees ßy from predawn until ßower closure, other potentially important pollinators 0.2 such as honey bees and bumble bees (Bombus spp.) were not active until well after dawn. Thus, we con- Þned insect surveys to a period from 0730 to 0900 0.1 hours (EST) to make sure that we would encounter all of Bees Per FlowerDensity the main pollinator species if they were present on the 0.0 target plant species at the study site. Pollinator species Honey Bees Bumble Bees Squash Bees that may show little activity before 0900 hours, such as sweat bees (Halictidae), are likely underestimated by Fig. 2. Density of three main squash and pumpkin pol- our methodology. Unless there is a shortage of polli- linators across 25 farms. Mean and SE given for each taxon nators, however, most of the pollinating activity has across all sites at which the taxon occurred (i.e., excluding zero values). already been carried out by that time of day. Surveys were carried out by one to three researchers trained to recognize the main pollinators without collection. Z-statistic as the original data ϩ 1 divided by the total Training was done through Þeld experience with an number of permutations ϩ 1. The algorithm was writ- experienced entomologist (T.H.R.) and through use ten in VisualBasic by T.H.R. of a local reference collection at Blandy Experimental We compared pollinator density (per ßower sur- Farm. Bumble bees were identiÞed only to genus. veyed) of honey bees, P. pruinosa, and bumble bees by At each farm, we recorded bee abundance and iden- using a general linear model analysis of variance tity by using methods based on those of previous (ANOVA) with farm and pollinator type as factors, researchers studying pumpkin pollination (Willis and pollinator density as the response variable, and the Kevan 1995). During each survey, researchers slowly adjusted sum of squares as the error term. We tested walked along farm rows counting the number of open the signiÞcance of farming practices on pollinator den- ßowers and the number and type of bee species seen sity through a general linear model by using the cat- in ßowers. On farms with small plantings, all open egorical variables of tillage/no-tillage and pesti- ßowers were surveyed once, but on larger plantings, cide/no pesticide as factors. The general linear model Þelds were divided into regions and subsampled. was carried out using Minitab, version 13.1 (Minitab, Analysis. Many participating farms were clustered Inc. 2000) statistical software. The density of honey geographically and any regional environmental fac- bees on squash and pumpkin on farms with and with- tors inßuencing pollinator populations could inßu- out managed honey bee colonies was examined using ence our results independent of farming practices. a t-test. Therefore, we tested for spatial autocorrelation as a prerequisite to the use of parametric statistics for pre- Results senting our results. Following Sokal and Rohlf (1995), we conducted a Mantel test of association between The specialist bee P. pruinosa occurred at 23 of 25 corresponding elements of two matrices. One of these sites and was the most common squash and pumpkin matrices held the intersite distances for all 25 partic- pollinator at 15 of those sites. Bumble bees visited the ipating farms based on latitude and longitude; the target plant at 16 sites, predominating at six, whereas other matrix held intersite differences in pollinator honey bees were observed at 13 sites and predomi- population density for all 25 farms. Separate matrices nated at four. In addition to visiting squash and pump- were generated for squash bee, bumble bee, and kin at more sites, P. pruinosa occurred at greater den- honey bee intersite differences in population size. We sities within sites (Fig. 2). Other sporadically calculated the Mantel Z statistic as the sum of the cross encountered bees that generally occurred in low products of corresponding elements of the two ma- abundance were various halictid species (Lasioglos- trices. The signiÞcance of the Z-statistic was tested by sum spp., Agapostemon spp., and Augochlorinae), and a Monte Carlo algorithm that randomly shufßed the Melissodes bimaculata Lepeletier. elements of one matrix then calculated a Z-statistic There was no evidence of spatial autocorrelation on based on shufßed data. Because we were concerned pollinator population size for the main pollinators. Of with a positive association between pollinator popu- 4,999 permutations of the pollinator intersite similarity lation size and spatial proximity, we looked for asso- matrix, 1,355 showed as great an association between ciations in the randomized data that showed as much intersite distance and intersite population size as the of a positive association between these factors as the original data for P. pruinosa (P ϭ 0.27). There also was association in the actual data. We carried out 4,999 no association between intersite distance and intersite permutations for each test and calculated the test population size for honey bees (P ϭ 0.47) or bumble statistic as the number of permutations with as high a bees (P ϭ 0.86). June 2005 SHULER ET AL.: SQUASH AND PUMPKIN POLLINATION 793

Table 1. Effect of pesticide use and tillage on P. pruinosa Table 2. Effect of pesticide use and tillage on bumble bee density on squash and pumpkin flowers density on squash and pumpkin flowers

Source df Seq SS Adj SS Adj MS FP Source df Seq SS Adj SS Adj MS FP Tillage or no-tillage 1 1.0382 1.0382 1.0382 6.09 0.022 Tillage or no-tillage 1 0.0001 0.0001 0.0001 0.01 0.941 Pesticide use 1 0.0602 0.1063 0.1063 0.62 0.438 Pesticide use 1 0.0001 0.0001 0.0001 0.00 0.960 Error 22 3.7485 3.7485 0.1704 Error 22 0.3547 0.3547 0.0161 Total 24 4.8468 Total 24 0.3548

General Linear Model ANOVA with tillage and pesticide use (yes/ General Linear Model ANOVA with tilling and pesticide use (yes/ no) as factors. no) as factors.

Squash bee population density was inßuenced by 1971, 1974; Willis and Kevan 1995). Squash bees and tillage practice but not by pesticide use (Table 1). honey bees seem to be equivalent pollinators of cu- Farms that practiced no-tillage agriculture had almost curbits in terms of initiating fruit production, but a three-fold increase in squash bee density (Fig. 3). squash bees visit ßowers more quickly (Tepedino Neither honey bee nor bumble bee population size 1981), more reliably, and disperse pollen over greater was associated with either of these variables (Table 2). distance to conspeciÞc stigmas than honey bees Honey bee population size on squash and pumpkin (Ordway et al. 1987). Female squash bees collect nec- was not associated with any measured variable, in- tar and pollen from Cucurbita, and males search for cluding the practice of keeping honey bee colonies on mates in the ßowers during the morning and then the farm. Eight farms kept honey bees on the property, crawl into a ßower as it closes and remain there all but these farms did not have a greater density of honey afternoon and night. Squash bees have expanded their bees on squash and pumpkin than did farms that did geographic range northward by expanding their host not keep honey bees (Fig. 4). This was true whether range from wild Cucurbita to cultivated Cucurbita and honey bee management was considered a categorical now occupy most of the continental United States into variable or a continuous variable weighted by the eastern Canada (Kevan et al. 1988). Attempts to in- number of hives and size of the farm. There were troduce squash bees to Hawaii to improve yields examples of farms that kept honey bees but received (Michelbacher et al. 1971) were unsuccessful. no squash or pumpkin visitation by them, and exam- ples of farms that did not keep honey bees but did The biology of P. pruinosa makes it difÞcult to man- receive visitation, either through feral colonies or age them in agricultural settings. The bee is a solitary managed colonies at other locations. species that excavates nests in the ground near its host plant. Nests are up to 46 cm in depth (Kevan et al. 1988), but most offspring are placed between 12 and Discussion 22 cm in depth (Mathewson 1968). The immature All of the farmers that we spoke with were aware of offspring lay dormant in the nest from late summer the need for insect pollination in their squash and until the following summer, when they complete de- pumpkin Þelds, but most assumed that they were de- velopment and emerge. Thus, they are difÞcult to pendent on managed honey bee colonies or wild acquire as immatures, susceptible to ground pertur- bumble bee populations for successful pollination. bations, and difÞcult to introduce in large numbers to Several had heard of P. pruinosa, but none knew that an agricultural setting. Still, there has been some suc- they occurred in their Þelds. The biology of P. pruinosa cess in initiating and promoting the populations of is well known, and their value to agriculture has long another ground-nesting bee species, the alkali bee, been recognized by pollination biologists (Hurd et al. Nomia melanderi Cockerell, which pollinates alfalfa

1.0 0.08

p = 0.032 p = 0.572 0.8 0.06

0.6

0.04 0.4

0.02

0.2 Bee Density Honey Squash bees per flower surveyed 0.0 0.00 Till No Till No Hives Hives Fig. 3. Effect of tilling practice on P. pruinosa den- Fig. 4. Honey bee density on squash and pumpkin at sity. farms with and without hives. 794 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 98, no. 3

(Medicago spp.) (Mayer and Johansen 2003) in the effect from any particular pesticide. Other studies northwestern United States. examining the broad use of particular have Although P. pruinosa is widespread and common, documented local population declines in bee species few studies have attempted to examine its distribution (Kevan and LaBerge 1979, Kevan et al. 1997). at a relatively local scale. Our Þnding them at 23 of 25 Honey bees are commonly used for squash and farms suggests that they are very abundant in this pumpkin pollination, but it is well known among bee- region and either persist on most farms or disperse keepers that they prefer many other plants over these well from other source populations. Our research sug- crops. Our study suggests that keeping honey bees gests that tillage inßuences P. pruinosa population size. may often make a minor, if any, contribution to squash Because the bees commonly nest in the Þelds, it seems and pumpkin pollination. Four participating farms likely that tillage could harm their offspring. The exact kept 10 or more colonies of honey bees. On all four of mechanism, however, is not clear. Some offspring may these farms, squash bees occurred in greater density occur within tillage depth (10Ð20 cm in this study) than honey bees, averaging 7 times their population and could be destroyed. Many offspring, however, are density. Honey bees may have been making an im- placed below tillage depth and would not be directly portant contribution to the pollination of other crops, injured. It could be that the collapsing of the tunnels but even those farms with the greatest investment in above the nest is sufÞcient to interfere with offspring pollinator management were relying primarily on wild emergence the following year. Evidence that tillage is pollinators on squash and pumpkin. a causative and not merely a correlative mechanism is The decline of honey bees has increased attention that the population density of pollinators that do not on alternative pollinators. Although much research commonly nest in agricultural Þelds (bumble bees and has focused on the development of pollinators to re- honey bees) was not affected by tillage practices. place the honey bee should it decline further (Strick- Bumble bees are excellent pollinators of many ag- ler and Cane 2003), it has led to an increasing real- ricultural plants, including some, such as tomatoes and ization that native bees already make substantial solanaceous peppers, that are poorly pollinated by contributions to agriculture. As future studies further honey bees. Because of their long active season, they examine the effect of agricultural practices on polli- require a continuous supply of ßowering plants across nator populations, it will be easier to examine both the the year to build up their colonies. Thus, farms with a ecological and economic trade-offs of those practices. diversity of crops ßowering continuously might be expected to develop larger or more colonies of bumble bees. Although there was a tendency for bumble bee Acknowledgments abundance on squash and pumpkin to be greater on We are grateful to owners and caretakers of all partici- more diverse farms, this relationship was not statisti- pating farms, whose cooperation made this study possible. cally signiÞcant. This could reßect our sampling re- We thank Blandy Experimental Farm for lodging and re- gime (we measured bumble bee density on only one search facilities. Comments from Quinn McFrederick and crop, not the whole farm) or a problem with geo- two anonymous reviewers helped improve this manuscript. graphic scale. Bumble bees can forage well beyond the We thank the National Science Foundation (DBI-0097505) edges of most farms (Osborne et al. 1999, Kreyer et al. for a grant in support of undergraduate research. 2004) and thus the diversity of crops within a farm could represent only a small portion of the foraging options within their range. A study that incorporates References Cited relative size and type of surrounding habitats (Kre- Allen-Wardell, G., P. Bernhardt, R. Bitner, A. Bu´ rquez, S. men et al. 2002, Steffan-Dewenter et al. 2002) as well Buchmann, J. Cane, P. A. Cox, V. Dalton, P. Feinsinger, as crop diversity and farm size simultaneously will M. Ingram, et al. 1998. The potential consequences of likely be a stronger analytical tool for examining the pollinator declines on the conservation of biodiversity factors that control bumble bee population size. and stability of food crop yields. Conserv. Biol. 12: 8Ð17. In contrast to Kremen et al. (2002), we did not Þnd Batra, S.W.T. 1995. Bees and pollination in our changing an association between pesticide use and native bee environment. Apidologie 26: 361Ð370. Chapman, T. W., A. P. Willsie, P. G. Kevan, and D. S. Willis. population size. There are several possible reasons for 1990. Fiberglass resin for determining nest architecture this discrepancy. First, Kremen et al. (2002)found a of ground nesting bees. J. Kans. Entomol. Soc. 63: 641Ð strong combined effect of pesticide use and isolation 643. from natural habitat, with isolation from natural hab- Delaplane, K. S., and D. F. Mayer. 2000. Crop pollination by itat being the strongest effect. Most farms in our study bees. CABI, Cambridge, United Kingdom. were located near natural habitat. Most farmers that Frankie, G. W., R. W. Thorp, L. E. Newstrom-Lloyd, M. A. did use pesticides in our study used them at times Rizzardi, J. F. Barthell, T. L. Griswold, J.-Y. Kim, and S. when pollinators were not active. Because squash and Kappagoda. 1998. Monitoring solitary bees in modiÞed pumpkin ßowers last only 1 d, there is unlikely to be wildland habitats: implications for bee ecology and con- servation. Environ. Entomol. 27: 1137Ð1148. a residual effect of pesticides sprayed on ßowers with Hurd, P. D., Jr., E. G. Linsley, and T. W. Whitaker. 1971. the next dayÕs ßoral visitors. Second, farmers used Squash and gourd bees (Peponapis, ) and the many different pesticides of varying toxicity to bees, origin of the cultivated Cucurbita. Evolution 25: 218Ð234. and we were unable to quantify their application. Hurd, P. D., Jr., E. G. Linsley, and A. E. Michelbacher. 1974. Thus, our study design was weak for detecting an Ecology of the squash and gourd bee, Peponapis pruinosa, June 2005 SHULER ET AL.: SQUASH AND PUMPKIN POLLINATION 795

on cultivated cucurbits in California (: Ordway, E., S. L. Buchmann, R. O. Kuehl, and C. W. Ship- Apoidea). Smith. Contrib. Zool. 168: 1Ð17. man. 1987. Pollen dispersal in Cucurbita foetidissima Kevan, P. G., and W. E. LaBerge. 1979. Demise and recov- (Cucurbitaceae) by bees of the genera Apis, Peponapis ery of native pollinator populations through pesticide use and Xenoglossa (Hymenoptera: , Anthophoridae). and some economic implications. In Proceedings of the J. Kans. Entomol. Soc. 60: 489Ð503. 4th International Symposium on Pollination. Maryland Osborne, J. L., S. J. Clark, R. J. Morris, I. H. Williams, J. R. Agric. Exp. Sta. Spec. Misc. Publ. 1: 489Ð508. Riley, A. D. Smith, D. R. Reynolds, and A. S. Edwards. Kevan, P. G., C. F. Greco, and S. Belaoussoff. 1997. Log- 1999. A landscape-scale study of bumble bee foraging normality of biodiversity and abundance in diagnosis and range and constancy, using harmonic radar. J. Appl. Ecol. measuring of ecosystemic health: pesticide stress on pol- 36: 519Ð533. linators on blueberry heaths. J. Appl. Ecol. 34: 1122Ð1136. Peng, Y. S., and M. E. Nasr. 1985. Detection of honeybee Kevan, P. G., N. A. Mohr, M. D. Offer, and J. R. Kemp. 1988. tracheal mites, Acarapis woodi, by simple staining tech- The squash and gourd bee, Peponapis pruinosa (Hyme- niques. J. Invertebr. Pathol. 46: 325Ð331. noptera: Anthophoridae) in Ontario, Canada. Proc. En- Robinson, W. S., R. Nowogrodzki, and R. A. Morse. 1989. tomol. Soc. Ont. 119: 9Ð15. The value of honey bees as pollinators of U.S. crops. Am. Kremen, C., and T. Ricketts. 2000. Global perspectives on Bee J. 129: 411Ð423. pollination disruptions. Conserv. Biol. 14: 1226Ð1228. Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles Kremen, C., N. M. Williams, and R. W. Thorp. 2002. Crop and practice of statistics in biological research. Freeman, New York. pollination from native bees at risk from agricultural in- Steffan-Dewenter, I., U. Munzenberg, C. Burger, C. Thies, tensiÞcation. Proc. Natl. Acad. Sci. U.S.A. 99: 16812Ð and T. Tscharntke. 2002. Scale-dependent effects of 16816. landscape context on three pollinator guilds. Ecology 83: Kreyer, D., A. Oed, K. Walther-Hellwig, and R. Frankl. 2004. 1421Ð1432. Are forests potential landscape barriers for foraging bum- Strickler, K., and J. H. Cane. 2003. For nonnative crops, blebees? Landscape scale experiments with Bombus ter- whence pollinators of the future? Entomological Society restris agg. and Bombus pascuorum (Hymenoptera, Api- of America, Lanham, MD. dae). Biol. Conserv. 116: 111Ð118. Tepedino, V. J. 1981. The pollination efÞciency of the Mathewson, J. A. 1968. Nest construction and life history of squash bee (Peponapis pruinosa) and the honey bee (Apis the eastern cucurbit bee, Peponapis pruinosa (Hymenop- mellifera) on summer squash (Cucurbita pepo). J. Kans. tera: Apoidea). J. Kans. Entomol. Soc. 41: 255Ð261. Entomol. Soc. 54: 359Ð377. Mayer, D. F., and C. A. Johansen. 2003. The rise and decline Weinberg, K. P., and G. Madel. 1985. The inßuence of the of Nomia melanderi (Hymenoptera: Halictidae) as a com- mite Varroa jacobsoni on the protein concentration and mercial pollinator for alfalfa seed, pp. 139Ð149. In K. the hemolymph volume of the brood of worker bees and Strickler and J. H. Cane [eds.], For nonnative crops, drones of the honeybee Apis mellifera. Apidologie 16: whence pollinators of the future? Entomological Society 421Ð436. of America, Lanham, MD. Westerkamp, C., and G. Gottsberger. 2000. Diversity pays Michelbacher, A. E., P.D.J. Hurd, and E. G. Linsley. 1971. in crop pollination. Crop Sci. 40: 1209Ð1222. Experimental introduction of squash bees (Peponapis)to Willis, D. S., and P. G. Kevan. 1995. Foraging dynamics of improve yields of squashes, gourds and pumpkins. Bee Peponapis pruinosa (Hymenoptera: Anthophoridae) on World 52: 156Ð166. pumpkin (Cucurbita pepo) in southern Ontario. Can. En- Michener, C. D. 2000. The bees of the world. John Hopkins tomol. 127: 167Ð175. University Press, Baltimore, MD. Minitab, Inc. 2000. Minitab, version 13.1. State College, PA. Received 20 May 2004; accepted 8 January 2005.